Analytics Model

Code

from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.preprocessing import LabelEncoder
from sklearn.metrics import adjusted_rand_score
eda = pd.read_parquet("data/eda.parquet")

Code

features = eda[['SALARY', 'MAX_YEARS_EXPERIENCE', 'MIN_YEARS_EXPERIENCE']].copy()

for col in ['MAX_YEARS_EXPERIENCE', 'MIN_YEARS_EXPERIENCE', 'SALARY']:
    features[col] = pd.to_numeric(features[col], errors='coerce')

features = features.dropna()

scaler = StandardScaler()
X = scaler.fit_transform(features)

kmeans = KMeans(n_clusters=4, random_state=688)
eda.loc[features.index, 'Cluster'] = kmeans.fit_predict(X)

true_labels = eda.loc[features.index, 'SOC_2021_4_NAME']
true_labels_encoded = LabelEncoder().fit_transform(true_labels)

ari = adjusted_rand_score(true_labels_encoded, eda.loc[features.index, 'Cluster'])

Code

import plotly.express as px
import plotly.graph_objects as go
from IPython.display import HTML
import numpy as np

# 1) Build the DataFrame
df_plot = features.copy()
df_plot['Cluster'] = eda.loc[features.index, 'Cluster']

# 1.1) Create a log salary column for better scaling
df_plot['Log_SALARY'] = np.log1p(df_plot['SALARY'])  # log(1+salary), avoid log(0)

# 2) Create an interactive Plotly Figure
fig = px.scatter(
    df_plot,
    x='Log_SALARY',  # 👈 use log salary instead of salary
    y='MAX_YEARS_EXPERIENCE',
    color='Cluster',
    title="KMeans Clustering by Log(Salary) and Max Years Experience",
    labels={
        'Log_SALARY': 'Log(Salary)',
        'MAX_YEARS_EXPERIENCE': 'Max Years Experience',
        'Cluster': 'Cluster'
    },
    width=800,
    height=600,
)

fig

Here we have 4 cluster groups. Group 0, which represent as green have lower salary, mostly under 150k, and max years experience in 2-5 years, it is likely Likely junior to mid-level employees with moderate pay. Group 1 with orange, has medium to high salary, wide range from $100k–$500k and with narrow range ~3 years, they are suggests specialized or high-paying roles with short experience — possibly fast-track promotions or high-demand fields. cluster 2 are low salary and experience from 0-4 years, they are clearly entry level employee. cluster 3 has medium salary, mostly under 200k with higher experiences, like 6-13 eyars. They probably are senior professionals with more experience but not the highest salaries.

Code

import pandas as pd
from sklearn.linear_model import LinearRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error, r2_score
import plotly.graph_objects as go

# Prepare features & target
features = eda[['MIN_YEARS_EXPERIENCE', 'MAX_YEARS_EXPERIENCE']].apply(pd.to_numeric, errors='coerce')
features = features.dropna()
X = features
y = eda.loc[X.index, 'SALARY']

# Train/test split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=688)

# Fit model & predict
model = LinearRegression()
model.fit(X_train, y_train)
y_pred = model.predict(X_test)

# Metrics (optional, but handy)
mse = mean_squared_error(y_test, y_pred)
r2 = r2_score(y_test, y_pred)
print(f"MSE: {mse:.2f}, R²: {r2:.3f}")

# Define min/max for the identity line
min_val = y_test.min()
max_val = y_test.max()

MSE: 758547543.20, R²: 0.096

Code

import numpy as np
import plotly.graph_objects as go

# Apply log transform
y_test_log = np.log1p(y_test)
y_pred_log = np.log1p(y_pred)

# Plot
fig = go.Figure([
    go.Scatter(
        x=y_test_log,
        y=y_pred_log,
        mode='markers',
        marker=dict(color='skyblue', opacity=0.6),
        name='Predicted vs Actual (Log Scale)'
    ),
    go.Scatter(
        x=[min(y_test_log), max(y_test_log)],
        y=[min(y_test_log), max(y_test_log)],
        mode='lines',
        line=dict(color='red', dash='dash'),
        name='Ideal Fit'
    )
])

fig.update_layout(
    autosize=True,
    height=400,
    title="Predicted vs Actual Salary (Log Scale)",
    xaxis_title="Actual Log(Salary)",
    yaxis_title="Predicted Log(Salary)",
    margin=dict(l=20, r=20, t=50, b=20)
)

fig.write_html(
    "figures/analytics_plot2.html",
    full_html=False,
    include_plotlyjs="cdn",
    config={"responsive": True}
)

fig.show()

This plot shows the Actual vs. Predicted Salary using a multiple linear regression model. The blue dots represent individual predictions, and the red dashed line is the ideal line where predicted = actual. Since most points lie very close to the red line, it means your model predicts salary very accurately, with minimal error and strong linear fit — likely reflected in a high R² score near 1.0.